Introduction to AI in Business - The 5th Industrial Revolution (Session 2)

• Session 1 - Embark on a journey through the basics of AI in simple terms. Dive into the 2025 AI trends and its integration into business operations, all supported by well-recognized and reputable sources.
• Session 3: Understand the critical aspects of AI accountability, safety, and risk management in business settings. Discover key frameworks such as ISO 42001 AI Management Systems, ISO 23894 AI Risk Management, and NIST AI Risk Management Framework 1.0. Uncover the compliance mandates of the EU AI Act of 2024.

Transcript

Welcome to this short course on artificial intelligence in business, the fifth Industrial Revolution. I'm Allen Keele. Thanks for joining me. We'll begin the course by getting a basic understanding of artificial intelligence in the business context. We'll then expand upon that by going into AI adoption and trends globally in 2024. We'll then look at how to benefit from and leverage AI in an organization. We'll then get more specific examples of how organizations are benefiting from integrating AI in various industry sectors. We'll then move on to managing AI risk, AI, accountability, safety and risk. We'll then learn more about how we can leverage existing newly emerging standardized frameworks to plan, integrate, manage and improve AI in business. And then finally, we'll go into late breaking newly emerging AI legislation and regulation that governs how AI can be used in business. 

Well, we've learned what AI is a bit about, how it works, where it's been, where it's going. Let's talk about how AI is actually integrated into your organization. How AI will change the way that you will conduct your business? Let's look at introducing AI into your organization. Now I've had customers that have told us, yeah, we're we're thinking about getting into AI and and seeing what to do with that probably late 2025, maybe Q 12026. Yeah. You know, when AI matures, folks, it's already here. 42% of companies surveyed already have integrated artificial intelligence into their processes. And here's the dilemma. That means that probably 58% of the companies that haven't have it, it's already been implemented through systems and software updates by their software vendors and suppliers. And they just don't even know it. 59% of companies surveyed have accelerated the introduction of artificial intelligence within the past 24 months, which only makes sense given how recent the development acceleration is. 32% of companies surveyed have. A holistic AI strategy in place that regulates the use of AI throughout the company. Making the most of what AI can bring in terms of competitive advantage in terms of efficiencies, but also. Managing the risk of AI, we've talked about some of this, the risk of potential breach of privacy, the potential breach of protecting personally identifiable information, PII. There's a lot of risk that comes with artificial intelligence. The bias that might have crept into our hiring process due to taking advantage of AI vetting our initial screening of applications. And doing so with bias that we weren't aware of. So it's not just about getting the benefits of AI, it's also about having a strategy to manage the risk of it. Going to be talking more about that in a moment. And then finally, we have 43% of companies surveyed report a positive impact on employee productivity. Due to AI and automation investments, we've already been automating, but now we need to have smart automation. The difference? Automating is static. Smart automation will change automated activities. Adapting to the scenario at hand, the time of day and the context of the activity that you've automated. That's the difference between regular rote automation and smart. AI enabled automation. Again, in talking with customers and we're trying to determine how to help them integrate AI into their organizations. A lot of times we can ask, for example. How critical is it that your employees get information and training on how to use prompts better for ChatGPT and other generative AI? That's. That's absolutely critical. You know, that's the training that we need. And I say, well, what about all of the AI that's happening without prompts? Are you ready for this? This is when I get the question. What do you mean? AI doesn't use prompts. You know, just this morning I was reading about the release of the new Samsung Galaxy S25 Ultra and how it's incorporating Gemini AI in the background. To make this phone a smart device that can go way beyond facilitating communications via telephone calls and text. To actually recognizing according to your schedule, what meetings you have and the topics at hand, and prepares a plan for you to better prepare for your meetings. It prepares an actual help with a sleep schedule to help you be better rested and prepared for upcoming business trips. In other words, it is considering information available to it that you didn't really give explicit consent to that you were aware of, and it's using this to optimize. Your daily schedule planning. Well, I mentioned this to one of my colleagues and she said, oh, this is fantastic. I love that. And I said, great, that's a great benefit. So how important is it that you learn, you know, prompts better for that? She said. Oh, that's critical. And I said it doesn't use prompts. I want to shake these people by their collars and point out the majority of AI that's in your business doesn't use prompts. It's already integrated into your software. It's running. Right now. It's helping with our accounting, our sales, our marketing, our HR management, quality management, supply chain management and business processes. And perhaps we could help it to work better. Perhaps there are features that we can enable that will allow it to do an even better job. Or perhaps you know? It turns out I just thought about it. If Gemini is already considering a lot of information. About me on the phone as to what it recommends that I do for future activities. That means it's accessing a lot of information. And I wonder. Well, who else is it sharing that with? And people say, well, of course nobody. Oh really? Did you make that assumption? Did you ask Jim and?
I.
Well, well, no. And if you did asked Gemini or Google, would they provide the transparancy? Talked about earlier. To let you understand what information is being considered, and moreover how that information is being used to actually make decisions about what it thinks you should be doing. Where's the transparency and how it thinks? So you can then. Trust its decisions and was that ever explained in a way that you could understand transparency and explainability? Told you that comes back and without transparency and explainability. We don't have AI trustworthiness, which is a a pinnacle goal for AI, is to achieve AI trustworthiness and ethical AI. So again, what I'm trying to do in this slide is to show you and break you out of the mold of thinking. This is just about generative AI. About creating stuff. In Dali, it's about creating. New emails it's about. Analyzing PDFs and getting summaries. It's far more than that, and it's already going on in these areas. O the most common uses of AI and generative AI in companies will use generative AI for creating first drafts of text and documents. We use it for personalizing marketing, for summarizing text, identifying trends and customer needs. We use it within chat bots. We use it to draw technical documents to help us with our coding. Example, we predict trends and fluctuations in business. We create first draft documents as I mentioned. We develop new products and designs, all purposely using generative AI and that's great. But again, even if you do feel good about what you're doing with generative AI, have you ever been disappointed in the product that it delivers and has it ever made you wonder? How did it ever get the idea? That I wanted these plants in my picture and that in other words, what was its thinking to come up with a picture that, Oh well, it all comes back to the prompts, Allen. OK. Since you know so much, tell me how it is considering those prompts. So I'll. It better be transparant with me and explain it in a way that I can understand so I can make better use of it. Can't really do that. There we. Now we're back to transparency and explainability and trustworthiness. O generative AI use cases and companies. These are coming from. Examples of companies that have deployed artificial intelligence in at least one area, and again this comes from an external source. This source Mackenzie, the state of AI in 2023. Generative Ai's break out. Year. So let's get a better feel for how AI is being deployed within specific functions. Now, before I get into this, please take note that this information is coming from 2021-2022, 2023-2024 numbers. Hard to come by. The percentage of respondents who stated that the specific AI function is used in at least one process or department. Typical response was robotic process automation leading the way. We're we're introducing again, not just having process automation, but smart process automation, computervision, natural language text understanding. Virtual agents or conversational interfaces, deep learning and the like. What about the number of business functions in which respondents have introduced artificial intelligence? Well. This ranges from 2021 through 2023, remaining relatively constant. However, introducing into five or more functions, we've had a significant drop since we first started considering AI in 2021. Counterintuitive. So far we've been looking at AI adoption and integration experimentation as kind of a binary question, meaning you either are or you aren't. Well, here's the dilemma. You might be adopting integrating experimenting with AI, but there are different levels. Of maturity in this some companies are doing it to a great extent, others not so much. So let's take a look at this. We have again those that have not adopted integrated. Or experimented with AI at all? That would be level 0 non automated companies. AI at all. We then have semi automated companies that is considered to have implemented weak AI such as what we've discussed. We then have level 2 automated companies incorporating strong AI that may incorporate some deep learning. And then we finally have Level 3 super intelligent companies. That are using AI to fully emulate an attempt to imitate human decision making and human interaction. In addition to our prerequisites for planning and integrating AI, we need to also. Anticipate what kinds of challenges and risks that we may encounter when adopting AI? What could cause adoption and usage of artificial intelligence to fail? Well, the very first one is. Our organization's lack of skills, we actually don't understand how to even set up a management system, a program for doing all of this work. By the way, you're going to understand later when I teach about ISO 42,142, double 01. Is a recently released ISO standard A framework giving us a road map, a template for establishing an AI function in an organization. To well manage the prerequisites for AI introduction to consider challenges and risks when adopting AI. What function is going to do these things? So it turns out the very first skill sets that we need to improve are those of the organization's governance. The people who make decisions about how the organization is going to be structured and run its overall strategic goals to even what AI is an executive governance decision to begin with. How are they prepared? With the right information and skills transfer to make good decisions on this. This is the kind of training that I'm roviding now that hels with that. But of course, like I said, more indeth training is going to be in understanding how we can leverage ISO 4201 AI management systems to establish an AI function in an organization. Understanding how we can use AI risk frameworks to manage the risks that automatically accompany the benefits of AI. All of this. Is competence and skills that need to be established and validated before we ever begin actually attempting to integrate AI. We also have to consider the potential high cost of AI, the cost associated with AI, the preparation it requires may be too high. Inadequate tools and platform using AI is challenging without the necessary tools or platform in which to develop models. We'll look at this. You'll have alternatives for this later, by the way. With AI as a service, for example. Something to. We'll talk more about that later. We have to consider the AI projects like your organization and its business can be quite complex. If you have a very complex business and organization introducing AI throughout, it is going to be a complex endeavor as well. When projects are too complicated, it can be very challenging to successfully integrate and scale AI systems. Also, we need to consider potential high data complexity when data complexity is high. AI deployment requires a lot of expertise specialists and significant computing power. That might go back to the previous high cost that I mentioned previously. AI doesn't make itself and it comes with expense that has to be budgeted. And then finally, confidence companies are struggling to really understand how to use AI responsibly and achieve the benefits of AI while managing and mitigating the risks of AI. So, Allen, I'm convinced we need to do better with integrating, adopting, experimenting with AI. We're ready to move forward and lead rather than fall behind. How do we get started? Well, let's look at some of the prerequisites for introducing AI integration and utilization. In other words, this is kind of like project planning. Figuring out what we need to have in place. To even be able to start planning and. Considering implementing integrating AI, the very first step is to better prepare an involve your existing employees. The dilemma is. They may not know what they need to know to really be able to help you in integrating. It's the blind leading the blind, as we say. So we have to understand weaknesses in the process. Using AI can improve processes. Problematic processes. Then need to be identified. In other words, you can't fix what you haven't identified as broken. So we need to identify what processes could be improved. By introducing artificial intelligence. We need a solid infrastructure, a platform to work from AI applications need reliable data exchange and a high performance infrastructure. We may not have that in place. We're we're getting things lined up to be able to do AI. This is one of the prerequisites we need to consider. The volume of data is going to increase significantly for AI to work autonomously, it requires a large amount of available data to learn from. After all, if it's going to make decisions like a person. It needs to get information first the knowledge. Before it can practice cognitive thinking to process that knowledge and produce. Actionable decisions. So we also had to be concerned about the quality of information is considering. Data cleaning high quality data helps AI applications. Make better decisions and perform tasks better as well. It only makes sense. Linking our data and processes data should align with our processes to ensure that AI is not taught incorrectly and also be considering again. Risk at hand, we want to make sure that AI has all the information and data that it needs, but not more. It's the principle of least privilege. It's information security, cybersecurity. We also have to consider our employee involvement. Employees need to be familiarized with AI and their new task well in advance. Need to be trained on how to work with AI in advance. These are considerations ahead of ever even planning and getting started with integrating AI into the organization. As we've seen, the first step towards planning, developing and integrating AI is to establish AI competence, especially with the very people who make decisions on even adopting and implementing AI to begin with. We'll need to get advanced training on traditional it as well as operations technology. And on the use of intelligent machines, we'll need perhaps to get some process consulting. We'll need training, perhaps on how to improve our particular business processes through AI. We'll also need to consider our platforms and data management. Employees will need new expertise in big data and cloud technology. Algorithms our technicians need to become skilled in developing AI algorithms. If we are in any way going to be able to customize our AI to our particular business needs and our particular business processes. And then finally leadership and. And perhaps this is where it should have all began. Changes in the division of Labor between humans and machines. Is going to. A big cultural change in your organization all the way from the leadership down. AI systems are essentially going to be new colleagues. And by the way, you've already hired some of these new colleagues that came into your systems and software through updates? And you didn't even get the chance to interview them. You don't even know. If you should trust them. But they're already on the job. Think about that. So going into more specifics, what about that leadership and competency development throughout the organization? Understand that intelligent machines will take care of a lot of our rote administrative tasks. There will be changes then in the division of Labor. Between humans and machines. Don't you think that it would be good to plan that change that's going to be a huge cultural change within the organization itself? Employees will need to develo problem solving skills and learn how to formulate questions in a way that machines can understand. Examples of how to better support our employees in using AI to increase productivity in the organization. They'll need to learn about how AI can. Well, automate repetitive tasks that will reduce employee workload, leaving employees to do what employees are needed for. Rather than what we can use AI for, we need to also consider potential error reduction in areas that are challenging and laborious. Nice thing about AI is AI will be consistent in how it works if we train it properly. And we'll be less prone to human error. We'll potentially have higher product quality through better error prevention because AI is going to be very diligent and on the job to the same level at all times. So long as it's been trained to do the job properly, of course we also have opportunities for new workforce that are specializing in AI to make all of these wonderful things happen. We will also need to consider personalized solution for employees as we migrate into. AI again? We're going to have traditional workforce. That is now not going to be the model moving forward. How will we help those people transition? We need to consider support designed to facilitate more efficient work. You know, oftentimes I know that with SAP implementations, for example, an SAP systems. I venture to say that probably most that have any experience there would agree with me. The first question you typically hear after trying to develop, implement and establish SAP is OK. This is how I used to do it. How do I do that with SAP now? And you have to explain, yeah, SAP doesn't do that. Oh well. But how do I? It to do what I used to do. You aren't going to do what you used to do. It turns out that was an inefficient process and SAP doesn't want to automate inefficient processes. It has an efficient process that you need to learn, and it's automated that. So at any rate, how do I do what I used to do? This is one of the issues about trying to progress. Your staff and employees and even management from how we used to do things to how we will do things in the future, so using AI. To enhance and increase productivity means doing things digitally and. I remember a movie that I once watched where gentleman his name was Garth and he said. People fear change. At any rate, point is, is that there's going to be a lot of human interface issues that we have to consider ahead of this. It doesn't just drop in and and plug in. Like perhaps the technologist would prefer. I can get more specific with some use case examples that go to specific functions in the organization, such as with hour. Now with hour we're already using some level of automation. Use automation, perhaps to collect data. About potential candidates for hire. We can also then automate the analysis of that data to some extent, and we're already sensitive to the fact that that automation can introduce errors. We can automate it to the point to where we end up considering. Some candidates based upon keyword searches and things of this nature that don't necessarily really map to what we need to accomplish. But then we can also end up inadvertently screening out excellent candidates because they didn't meet our. Our screening rules, OK. Well then from the information that we collected and that we've done some preliminary analysis on at some point, then we need to take that information. Consider it. And draw some insights from it and this is where AI can help. Many companies already collect and analyze data using automation. But concrete insights and knowledge are still gathered by humans. So this again is a is a potential that we could identify for introducing AI to improve efficiency and quality of process. There is the potential for decisions made during applicant selection. To not only be slow, but biased, artificial intelligence can automatically generate these insights and findings. Avoiding these issues mentioned and provide a better basis for decision making. So this is where. Could either fully or partially replace humans in this analysis. And insight development. This is where for example, we could use AI to help us make better decisions. About our candidates for hire. So I think you'd agree it was nice to take a hands on look at how AI could be integrated into a typical HR process of screening and hiring new candidates. That's pretty well applicable to, well, all organizations. All we hire. And. We have to go through this process. Let's go ahead and take a look at some industry sector use case examples of how we can use AI to improve our business to make us more competitive. To reduce costs in various industry sectors. We'll begin by looking at integrating AI into the healthcare sector. Now, even though it's healthcare and you may not be in the healthcare sector, understand a lot of what I'm going to show you could be adapted to your particular industry sector as well. So let's take a look in the healthcare sector. We could use a machine learning program to remotely analyze patients health via a mobile device, comparing it to medical records and warning of potential diseases. This is already being used. In doctor's offices today, obviously. And it improves quality of care through AI. We also have AI that could use image recognition during an examination. For example, comparing the collected data with specific disease patterns, flagging conspicuous areas on X-rays and CT scans. And again, it will augment a physician compliment a physician helping them. Not miss symptoms. Helping them to do a better job of being a better physician. But they're still the physician with the assistance of AI not being replaced by AI. We have autonomous diagnostic services that could perform simple medical tests without human assistance leaving. Valuable resource in our physicians and nurses and nurse practitioners to do what AI cannot that relieves the doctors and nurses of routine activity. It can also help and assist by predicting patient behavior. Likelihood of illnesses that are not yet. AI algorithms can optimize hospital processes, staff schedules, inventories, reducing costs, increasing quality of care. AI tools could also be used to analyze environmental factors in patients medical backgrounds to recommend preventive care programs, virtual consultants register patients. Recommending suitable doctors, reducing weight time. Performing a type of triage, if you will. We have AI personalized treatment programs that improve treatment efficiency by tailoring the treatment to the patients, particular medical needs and conditions, and then finally, AI could perform compare large population health analysis. Helping us identify trends that we may have otherwise missed and doing it in more current real time. Let's move to integrating AI into the retail sector. We could use machine learning and facial and speech recognition to enable virtual advisors to greet customers in person, predict their purchases. And perhaps point them in the right direction. Now there is already some controversy about this because we now have department stores, for example, that have electronic signs, billboards, if you will, their. Integrated throughout the store. That through the use of facial recognition or actually tracking customers as they move through the store. So if you've already perhaps picked up a shirt or a blouse that you want to purchase even prior to buying it on the way to the register through another department, you might get an advertisement. For an accessory that goes with that particular blouse or shirt. So in other words, the advertisements are being tailored to you based upon what you were looking at justice a moment ago. And it's tracking you by your face. Well, great for the. You however, might perceive that as going a bit too far. Again, risk a companies. We'll talk more about that later, but in terms of how to use. AI to do a better job in helping the customer purchase better, we might call that sales, but to help the customer purchase better, we could implement. Artificial intelligence to draw the customer's attention to things that they would probably like to have, along with things they've already identified that they want. We can also recognize that machine learning could then use customer profiles again to personalize advertising, as I explained, as customers walk through the store, offers are sent to their smart phones, for example. A computer vision system identifying items packed by the customer at Self Checkout as payment is made automatically. The customer could then leave the store directly. Improving the purchase process and then finally, autonomous drones that use deep learning could take over short haul deliveries, reducing wait times and freeing up delivery crews. So again, these are just a few examples. Are hundreds of ways. That we can use AI to not only shortcut tedious, monotonous rote things that we've already been doing. AI can be used to accomplish new feats in marketing and sales that we've never even considered before. This is exciting and again think about it. If you're in the retail sector and your competitors are doing this. How can you hope to compete with that if you don't? So it's not a matter of if you're going to adopt AI. You don't. It won't matter anyway. Won't be able to compete. How do we do that? Best that's what we're going to look at soon. We continue in integrating AI into the retail sector with interactive. They could identify items recommending complimentary products based upon the customer's profile. As I've already talked about, we could have autonomous shopping carts following the customer in the store automatically finding their way to the car. Or a drone if it's for home delivery. We have stores that could optimize their prices in real time, maximizing their profits. This again has met with a lot of criticism and has become quite contentious. There is a very well recognized, well known retail grocery store chain that has started to implement this by recognizing. Based upon your purchase patterns in one aisle, maybe you're the person who gravitates towards purchasing quality. Without necessary regard for expense, you would prefer to spend more to get better. Well, it turns out that you don't know it, but as you progress to other aisles. They'll dynamically change the price as you approach, knowing that you're willing to spend more. Well, that's not fair. But it could help them maximize their profits. Couldn't. So again, this is just going to show is AI good or bad. I guess it's in the eyes of the beholder now, isn't it? We also have robots with artificial intelligence that again can also continuously track inventory, detecting empty shelves, refilling those empty shelves, all of the rote processes that we had people doing, we can now have done more efficiently with more accuracy at less cost. Using AI and robotic replacement and inventory replenishment. So this is. Yeah, let's take a look at integrating AI and energy supply. And now we're getting into more of an industrial application, right? We have sensors, a machine. We've already had sensors, but now we can enable real time response to wind conditions, perhaps maximizing green energy production. We can use machine learning to enable forecasting of peak demand, maximizing the use of short term renewable energy. We can use digital power systems to detect the current situation, adjusting the real time power supply. To a building's current grid load, we could use drones and robots to better predict failures and control equipment without interrupting production. I think we're starting to get the hang of it, so let's pick up the pace looking at integrating AI in the service sector, AI could automatically log and forward documents, leaving technical staff more time to focus on solving their own problems. We could have field staff receiving real time updates to shorten response times and reduce the impact of outages. We have virtual consultants that could automate call centers and segment customers. They could then resolve simple common customer issues on their own, or refer them to employee if necessary. We have a smart meter data and AI service providers that could offer new services. Based on usage patterns, weather and other factors. OK, I like where this is. Let's take it up a notch or 10. Let's talk about integrating AI with robotics. How can we get intelligent robotics to replace? Repetitive human tasks. Well, we need to train robotics to do so. And humans provide that. O this is where we begin with a human instructor guiding the robot as it learns and corrects wrong actions and movements as they occur. Again, it's going back to that learning process that I told you about earlier. It does have use. About learning through punishment and reward. Recognizing wrong actions, incorrect decisions. Punishment and also recognizing correct actions that are not then corrected because they were wrong. It recognizes those rewards and it learns through this process and gets better and better as it goes. So these trained robots can work together with human workers to, well, start replacing repetitive tasks, handing the human objects, for example, they can continuously improve their algorithm in the process. Again, through this punishment reward process. Once the robot has finished learning, perhaps it's gone so far as to get beyond just simply handing a human objects to actually managing the object entirely. It repeats the new routine of movements on its own and automatically completes the repetitive procedures. As necessary as it's been trained to do. And then finally, how do we keep an eye on all of this and make sure that it works properly? This is where we can have a control in place, sensors and cameras that provide information to computer vision algorithms, ensuring reliable cooperation between. Workers and robots. And all of this might lead us to consider how we could use AI and robotics to increase yield with artificial intelligence in the manufacturing sector, for example, we could use AI powered root cause analysis. To improve quality and problem solving, we could use an AI based analysis of process information to better predict yield disruptions and to help workers to better manage them. We could use AI to analyze data from AST Roduction runs and to help ensure greater schedule adherence in future processes. And finally we can use data on production tools. That is automatically transmitted to AI to determine optimal process conditions. We referenced using AI to managing quality and manufacturing and yield production and improvement. So let's drop down more into that quality management. The AI system can process thousands of images. To help us to automatically detect defects, it can detect defects. Far better than what we could do manually with humans just because of the sheer volume of what it can consider. Workers could then be automatically alerted to specific locations of a. So it can be rectified, remedied earlier rather than later, which, if you understand quality management, that's how we save money. And then finally, we could also have a list of the most common errors that could be created by analyzing pass errors in production. And again, we can have numerous cameras to take pictures that they intend to send AI machines for. Further analysis for further improvements and predictions. Continuing in quality management, we can have sensors that could detect sounds or vibrations. Sending the data to the AI system for processing. We have maintenance workers who could receive automatic suggestions for anticipated maintenance and scheduling. Machine learning algorithms can help us to accurately predict the maintenance needs of machine parts. And to better describe what needs to be done. And then finally, this predictive maintenance significantly reduces machine downtime caused by maintenance work. And we'll finish our use case examples with AI in supply chain management. AI can combine relevant internal and external data for highly accurate demand forecasting. We also have material flows and volumes that are automatically adjusted based on real time data such as weather conditions. Artificial intelligence is transforming the traditional supply chain into a full delivery network with a digital core. And all of this will provide for more accurate forecasting, allowing for lower inventory levels, which reduces costs both with acquisition as well as storage and insurance throughout the supply chain. Which means drastically reducing potential cost of product. This is fantastic. Now one last parting question before we leave our use case examples. We have covered 11 slides. Of 44 use case examples. At what point did I ever mention using ChatGPT, or generative AI? At what point did I mention? Understanding prompts better. I didn't. I only wanted to call this out because I keep facing it over and over again with. Business customers who still think that AI is chat. That think integrating AI into business is using ChatGPT to help us construct documents and do other generative AI tasks. It is, yes. But look at what we've just covered over these 44 use case examples. It makes it makes the other pale by comparison in terms of value. This is the magic of integrating AI into business, and This is why your organization can come out on top with it or disappear without it. It's as easy as that.